Holding the tension between an AI-forward brand and users who prefer AI unlabeled.
Two things are true at the same time inside the company right now, and they point in opposite directions.
The first: we’re building an AI-forward identity. A dedicated color for AI features. A unifying pattern for our AI assistants. A roadmap that treats AI as the headline.
The second: our own research keeps finding that when we put an AI label on a feature, engagement drops. Take the label off and engagement goes up. Users are skeptical of “AI” as a badge. They’re happy to use the capability. They’re wary of being told it’s AI doing it.
The easy move is to pick a side. Either users are wrong and will come around, so keep branding it. Or users are right, so strip the labels and ship AI invisibly. I don’t think either is the answer, and I think the contradiction is worth holding open a little longer before we collapse it.
Here’s the reframe that’s helped me. “AI” isn’t one thing to the user. It’s two. There’s AI as a capability, the feature that drafts the thing, catches the error, saves the step. And there’s AI as a claim, the word on the button announcing a machine did it. People are voting for the capability and against the claim. The label adds risk (is it accurate, can I trust it, what happens when it’s wrong) without adding value. The capability earns trust by working.
Which suggests the real question isn’t “should we brand our AI or hide it.” It’s “where does naming the AI help the user, and where does it just transfer our enthusiasm onto them as doubt.” A back-office task the customer never sees should probably stay invisible: show the result, not the machinery. A feature where the user has to calibrate trust, decide how much to lean on a suggestion, might genuinely benefit from the label. The brand identity and the research aren’t actually in conflict. They’re answering different questions. We’ve just been treating “AI” as a single decision when it’s a hundred small ones.